Patients' Melanoma Cells Engineered to Secrete GM-CSF

ASCO--Injections of an irradiated autologous melanoma vaccine geneticallyengineered to secrete GM-CSF proved safe and capable of eliciting an antitumorresponse in patients with advanced tumor burden, Robert Soiffer, MD, ofthe Dana-Farber Cancer Institute and Massachusetts General Hospital, saidat an ASCO scientific session.

Of the 20 evaluable patients, four had a small decrease in the sizeof tumor deposits, and some patients with no observed decrease in tumorsize had extensive tumor necrosis on pathologic exam. "This mightsuggest that perhaps we need to rethink how we measure response in thisparticular circumstance," he said.

At a median of 14 months' follow-up, 14 patients remain alive. Fourpatients with stable or responding disease are currently being re-treatedwith another vaccination course. "Three patients who have been renderedsurgically NED are not being treated and have been free of disease forover two years," he said.

The phase I trial was conducted by Drs. Soiffer, Thomas Lynch, and GlennDranoff, and their colleagues in collaboration with Somatix Therapy Corporation,Alameda, Calif.

The study enrolled 32 stage IV metastatic melanoma patients. The sourceof the vaccine was generally soft tissue mass or lymph node (22 patients),though tumor was also removed from the lung, liver, and adrenal gland.Harvested melanoma cells were expanded in culture and transduced with areplication-defective retrovirus expressing the GM-CSF gene.

Of 32 tumors harvested, 28 vaccines were successfully prepared, andthese cells secreted a median of 280 ng/million cells of GM-CSF over 24hours. Of these 28 patients, 25 actually received vaccination.

The interval between harvest and vaccination was 10 weeks (range, 8to 32 weeks), "but almost all received their first vaccination between8 and 13 weeks," Dr. Soiffer said. Five patients withdrew early becauseof progressive disease that occurred primarily during this harvest-vaccinationinterval.

Patients received 107 irradiated tumor cells monthly for a total ofthree injections, every 14 days for a total of six injections, or weeklyfor a total of 12 injections. Dr. Soiffer noted that the frequency of vaccinationmay be an important variable because the intensity of the pathologic reactions(described below) was greater at skin vaccination sites with more frequentinjections.

There was little toxicity, he said, except for local skin erythema andinduration at the injection site.

A number of patients with subcutaneous metastasis developed significantclinical signs of inflammation around these tumor deposits. "Whenwe looked at these specimens histologically, we found a significant rimof fibrosis with underlying central tumor necrosis," he said. "Whenwe looked more carefully, we saw lymphocytes invading and destroying bloodvessels that were supplying the tumor, as well as melanoma cells surroundedby lymphocytes."

These tumor-infiltrating lymphocytes were extracted from these tumordeposits, and their ability to kill the patients' own melanoma cells wasevaluated in the lab. In one patient, the T cells were able to specificallylyse her own tumor but did not lyse another patient's tumor, suggestingsome degree of specificity.

Serologic changes were also observed. "Prior to vaccination, therewas very little reactivity with melanoma cell panels, but after vaccination,we saw an increase in antibody production that reacted with patients' melanomacells," he said.

The clinical significance of these vaccine-induced responses needs tobe evaluated in future efficacy studies, Dr. Soiffer said. The researchersare planning ultimately to look at patients with a smaller tumor burden,perhaps in an adjuvant setting, and are examining new adenoviral vectorsin hopes of increasing the efficiency of gene insertion and thus decreasingthe interval between tumor harvest and initiation of therapy.